Influence of convective organization on the triple oxygen isotope composition of West African monsoon rainfall: insights from a 7-year record in Benin
Abstract. Triple oxygen isotopes, and specifically 17O-excess, provide new opportunities to investigate kinetic fractionation associated with atmospheric moisture sources, continental moisture recycling, and cloud microphysical processes. However, observations of 17O-excess in precipitation remain extremely scarce in Africa, leaving its potential for tracing atmospheric processes in the West African Monsoon (WAM) largely unexplored. Here we analyze a high-frequency precipitation isotope dataset (δ18O, δ2H, d-excess, and 17O-excess) from Benin, comprising a 7-year record at the inland Nalohou station and a complementary 2-year record from a coastal station near Cotonou. Three isotopic clusters are identified inland, corresponding to distinct rainfall regimes. The first is associated with short-lived convective events affected by sub-cloud rain evaporation, while the second reflects more organized convective systems and a weak signature of continental moisture recycling. The third cluster, observed only inland, is characterized by strongly depleted δ18O together with simultaneous decreases in d-excess and 17O-excess, and occurs mainly during the mature monsoon stage. Its rainfall characteristics are consistent with larger and more organized convective systems, with a greater contribution from their stratiform component. At the interannual scale, δ18O variability is more closely related to rainfall characteristics during the mature monsoon phase than to rainfall amount. Overall, these results indicate that, during the WAM mature monsoon phase, precipitation isotopes primarily reflect convective organization and associated cloud microphysical processes, while retaining a weak signature of continental moisture recycling during inland transport. These combined δ18O, d-excess, and 17O-excess data provide new observational constraints for isotope-enabled atmospheric and hydrological models.